People ask me constantly: how do you make so much content without burning out?
The answer isn’t “I work less.” It’s that I’ve built a system where each piece of content feeds the next, and AI handles the heavy lifting between pieces. One idea becomes three outputs. That’s the leverage.
Here’s the full workflow.
Phase 1: Idea Capture (Continuous)
Ideas don’t come when you’re sitting at your desk trying to have them. They come when you’re in the shower, watching someone else’s content, in the middle of a conversation.
I capture everything in a Notion database — voice memos, quick notes, screenshots. No editing at this stage. It’s just a place for everything to land.
At the end of each week, I spend 20 minutes reviewing that database and picking the 2-3 ideas worth developing.
Phase 2: Research and Angle (1-2 hours)
For each idea I move forward with:
Perplexity → I run 3-5 searches to understand the current conversation around this topic. What do people believe? What’s contested? What’s new?
ChatGPT → I ask it to give me the five most interesting things about this topic that most people don’t know. I don’t use the answers — I use them to check my angle. If ChatGPT’s first hit covers my angle, I need a better angle.
Personal experience → This is the most important part and it’s the one you can’t shortcut. What do I actually think about this? What have I personally tested? The AI can research — it can’t have an opinion that’s worth anything.
I take notes in a Notion doc throughout. By the end, I have a clear thesis and 3-5 supporting points.
Phase 3: TikTok Script (30 minutes)
TikToks are a different writing discipline — short, punchy, front-loaded. The hook has to work in the first 2 seconds or I’ve lost the viewer.
I write scripts by hand, in my voice. AI doesn’t write my scripts. What AI does:
- Claude → I paste my points and ask “what’s the most counterintuitive way to frame this?” I often don’t use the answer but it helps me find a sharper angle.
- Structure check → I paste my draft and ask “where does this get slow?” Useful for tightening.
Final script: almost always under 200 words. Then I film.
Phase 4: Filming and Editing (1-2 hours per video)
I film in my studio setup — good light, good mic, nothing complicated. Multiple takes on the hook, usually one or two takes on the rest.
Editing in CapCut. I’m not a professional editor and I don’t pretend to be — clean cuts, captions, maybe a sound effect. The content carries it.
Runway → If I need B-roll I don’t have, I generate it. Usually 10-20 seconds of supporting visuals.
Phase 5: Blog Post (45-60 minutes)
This is where the system gets efficient.
After a TikTok is done, I have:
- A researched thesis
- Supporting points with evidence
- My own perspective on the topic
That’s most of a blog post. I expand the TikTok into a 600-1000 word piece — same angle, more depth, more nuance.
Claude helps me structure the expansion:
“Here’s my TikTok script about [topic]. Expand this into a long-form blog post that goes deeper on each point. Keep the voice conversational.”
I rewrite the output significantly, but it gives me a scaffold to work against.
Phase 6: Newsletter (30 minutes)
The newsletter has a fixed format (I covered this in another post). For the main review section, the blog post gives me the content — I summarize it into newsletter format, add the quick hits and the editorial take, and it’s done.
Three pieces of content. One research process.
The Tools in Order of Use
| Phase | Tool |
|---|---|
| Idea capture | Notion voice memos |
| Research | Perplexity, ChatGPT |
| Script | Written by hand, Claude for feedback |
| Filming | CapCut, Runway for B-roll |
| Blog | Claude for scaffold, I rewrite |
| Newsletter | Written from blog summary |
What the System Doesn’t Fix
Consistency. You still have to show up. The best workflow in the world doesn’t work if you don’t ship. There are weeks where I don’t feel like creating anything, and I do it anyway.
That’s not an AI problem. That’s a human problem. And the only solution is still showing up.
— Deco